Simulated use of thresholds for precautionary allergen labeling: Impact on prevalence and risk

Heterogeneity and overuse of precautionary allergen labelling (PAL) in prepackaged foods have eroded its risk communication efficacy. Experts recommend applying PAL based on allergen concentration thresholds, but adoption remains limited. The aim of this study was to quantitatively assess the potential impact of this approach using Monte Carlo risk simulations. Four allergens and 9 food categories were considered in 2 scenarios: (1) consumption of products currently carrying PAL in Canada where individuals with food allergy (FA) are assumed to consume them, and (2) consumption of products without PAL, in a hypothetical context where PAL is applied based on thresholds that would protect 99 % (ED01) and 95 % (ED05) of individuals with FA, and individuals with FA systematically avoid products with PAL. In scenario (1), although several cases studied would cause <10 reactions/10 000 eating occasions (e.o.), there were also many that would cause >20 reactions/10 000 e.o. Cross-contact milk posed the highest risk (max. 1120 reactions/10 000 e.o.), and peanut, the least (max. 10 reactions/10 000 e.o.). In scenario (2), consumption of products without PAL, when using thresholds for PAL based on ED01, could lead to a maximum of 15 reactions/10 000 e. o. for all studied cases, and based on ED05, to 57 (if excluding dark chocolate with milk PAL). In most cases, the estimated number of reactions per 10 000 e.o. attributed to products with PAL currently on the market would be higher (p < 0.05) than that attributed to products without PAL, if PAL is applied based on the simulated thresholds. Thus, a threshold driven approach to adopt PAL on prepackaged foods, while advising consumers to avoid these products, could be beneficial for individuals with FA in Canada, as products without PAL would result in very few and generally mild adverse reactions.